Graph-Slam(一)_概率机器人graphslam-CSDN博客

本文链接：https://blog.csdn.net/xdgs_2005/article/details/51706574

GraphSLAM是一种将离线SLAM问题统一的算法，通过将SLAM后验概率转化为图网络，利用变量消除技术降低问题维度，然后用传统优化方法求解。这种算法能生成包含108个以上特征的地图。内容主要涉及SLAM问题的离线形式，机器人在不同时间的位姿表示，地图的定义以及测量和传感器的功能。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

GraphSLAM is a unifying algorithm for the offline
SLAM problem.It transforms the SLAM posterior into a graphical network,
representing the log-likelihood of the data. It then reduces this
graph using variable elimination techniques, arriving at a lowerdimensional
problems that is then solved using conventional optimization
techniques. As a result, GraphSLAM can generate maps
with $10^8$ or more features.
1 Mapping SLAM Problems into Graphs
1.1 The Ofﬂine SLAM Problem
In SLAM, time is usually discrete, and t labels the time index. The robot pose at time t is denoted $x_t$ ;
we will use $x_{1:t}$ to denote the set of poses from time 1 all the way to time t. The world itself is denoted m, where m is short for map. The map is assumed to be time-invariant, hence we do not use a time index to denote the map. In this paper, we think of the map of a (large) set of features $m_j$ .
To acquire an environment map, the robot is able to sense. The measurement at time t is denoted $z_t$ . Usually, the robot can sense multiple features at each point in time; hence each individual measurement beam is denoted ${z_t}^i$ . Commonly, one assumes that ${z_t}^i$ is a range measurement. The measurement function h describes how such a measurement comes into being: